JOURNAL OF VIROLOGY, Feb. 1991, p. 796-804 Vol. 65, No. 20022-538X/91/020796-09$02.00/0Copyright ©) 1991, American Society for Microbiology

Inverse Relationship between Human Papillomavirus (HPV) Type 16Early Gene Expression and Cell Differentiation in Nude Mouse

Epithelial Cysts and Tumors Induced by HPV-PositiveHuman Cell Lines

MATTHIAS DURST,l* FRANZ X. BOSCH,2 DAGMAR GLITZ,' ACHIM SCHNEIDER,3AND HARALD ZUR HAUSENI

Institut fur Virusforschung, Deutsches Krebsforschungszentrum, Im Neuenheimer Feld 280,1 and Hals-Nasen-Ohren-Klinik, Universitat Heidelberg,2 6900 Heidelberg, and Frauenklinik, Universitat Ulm,

7900 Ulm,3 Federal Republic of Germany

Received 17 July 1990/Accepted 29 October 1990

Two human papillomavirus type 16 (HPV 16)-immortalized human keratinocyte cell lines (HPK) were shownto have retained the ability for differentiation after subcutaneous injection into nude mice. These propertieswere maintained even at late passage. HPK cells gave rise to transiently growing cysts which exhibited anepitheliumlike architecture. Moreover, differentiation-specific markers such as cytokeratin 10, involucrin, andfilaggrin were shown to be expressed in an ordered succession. RNA-RNA in situ hybridization revealedheterogeneous and low levels ofHPV 16 E6-E7 RNA in the basal layer of the cysts. In contrast, in progressivelygrowing tumors induced by HPK cells containing an activated ras oncogene (EJ-ras) or in tumors induced bythe cervical carcinoma cell line CaSki, high levels of E6-E7-specific RNA could be detected. Irrespective of thegrowth potential of these cell lines in nude mice, viral transcription was always more evident in the basal layerand in proliferatively active cells rather than in differentiated cells. This contrasts with viral gene expres-sion in HPV 16 positive low-grade cervical dysplasia, in which abundant viral transcriptional activity wasmapped to the upper third of the epithelium. It is suggested that the physical state of the viral DNA, i.e.,integrated viral DNA in the cell lines as opposed to extrachromosomal DNA in low-grade cervical dysplasia,may influence viral gene regulation.

To date, more than 60 different human papillomavirus(HPV) genotypes are known. A subset of these viruses isfound in the urogenital tract. HPV types 6 and 11 (HPV 6 andHPV 11) induce preferentially benign proliferative changesof the epithelium (condylomata acuminata), while otherssuch as HPV types 16, 18, 31, 33, and 35 are associated withflat condyloma, precancerous lesions (squamous intraepithe-lial neoplasia), and squamous cell carcinoma.The assumption that certain papillomaviruses play an

active role in the development of cervical cancer is based notonly on the detection of viral DNA in most premalignant andmalignant lesions but also on, among other findings, theability of viral DNA to alter the growth behavior of humancells in vitro (for a review, see reference 36). Recent studieshave shown that cultivated human primary foreskin kerati-nocytes became immortalized in response to transfectionwith HPV 16 DNA (11, 25). Similar results were observedfor HPV 18 and other cervical cancer-linked HPV types butnot for HPV 6 or HPV 11 (24, 29, 33). The immortalizingfunction could be mapped to the viral genes E6 and E7 (15,22). The observation that the HPV 16 E7 protein forms aspecific complex with the human retinoblastoma suppressorgene product p105-RB (13, 23) may be relevant for themechanism by which these viruses contribute to cell trans-formation. Similarly, the E6 protein has been shown to bindp53 (32), another protein suspected to play a role in tumorsuppression. None of the aforementioned HPV-immortal-

* Corresponding author.

ized human keratinocyte cell lines were tumorigenic in nudemice. However, the tumorigenic phenotype could be in-duced in these cells by the cooperating effect of an activatedras oncogene (10, 12).

In natural hosts, HPV infection is morphologically char-acterized by epithelial hyperplasia, with cells exhibiting anumber of abnormalities such as nuclear changes (enlarge-ment, binucleation, and degeneration) and cytoplasmic vac-uolization (koilocytosis). In cervical intraepithelial neoplasia(CIN), histological grading is based on epithelial criteriasuch as the proportion of the epithelial thickness occupiedby undifferentiated basal layer-like cells and on single-cellcriteria such as nuclear abnormalities (e.g., an increasednuclear-cytoplasmic ratio) and the presence of atypical mi-totic figures (6, 19, 28). Morphological alterations similar toCIN have recently been described for HPV-immortalizedkeratinocytes. By using a raft system which permits epithe-lial cells to stratify and differentiate in vitro, morphologicalchanges which closely resembled those seen in histologicalsections of low-grade CIN were observed for HPV 16- andHPV 18-transfected primary keratinocytes at low passage.Late passages of the same cells had completely lost theability to differentiate in the raft system, thus resemblinghigh-grade CIN (16, 18).

In this article, we show that immortalization in vitro ofhuman epidermal keratinocytes is not coupled with a loss ofterminal differentiation when the cells are grown under invivo conditions. Subcutaneous injection of two HPV 16-immortalized keratinocytes cell lines at high passage (>2years in culture) into nude mice gave rise to encapsulated

796

on February 5, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

DIFFERENTIATION-DEPENDENT HPV GENE EXPRESSION 797

cysts showing an orderly structured and differentiated epi-dermal tissue. Moreover, in these cysts, as well as in tumorsinduced by a malignant derivative of one of the immortalizedcell lines, HPV early gene expression appears to be inverselyrelated to terminal differentiation, which is marked byexpression of cytokeratins 1 and 10 as well as involucrin andfilaggrin (9, 21, 27). These observations contrast with thetranscriptional activity of HPV 16 in low-grade CIN (8, 31).We suspect that these differences in viral gene expressionmay in part be attributed to the physical state of viral DNAwithin the host cell. In CINs the viral DNA predominates inepisomal forms, whereas in cell lines established by trans-fection of HPV DNA the viral genome always persists in anintegrated state.

MATERIALS AND METHODS

Cell lines. The HPK IA and HPK II cell lines wereestablished by transfection of human primary foreskin kera-tinocyte cultures with the entire HPV 16 genome and sub-sequent selection for unlimited growth in vitro (11). At alater stage, the cell lines were adapted to grow in Dulbeccomodified Eagle medium supplemented with 5% fetal calfserum only. They have been kept in culture continuously forover 2 years, being passaged on average once per week(initially at a split ratio of 1:2 and later at a ratio of 1:8).During this time, the HPV 16 DNA integration pattern wasnot modified and the viral genome remained transcriptionallyactive. HPK IA cells harbor two to three complete copies ofthe HPV 16 genome arranged in a head-to-tail fashion,integrated at one chromosomal site. HPK II cells possessabout 10 copies of the viral genome, also arranged in ahead-to-tail fashion but integrated at two distinct sites withinthe host DNA. The precise locations of the integration siteswithin the viral genome of both cell lines have not yet beendetermined. Southern blot analysis of HPK IA DNA indi-cates that both virus-cell junctions are within the Li openreading frame (11). HPK II cells displaying a malignantphenotype were established after transfection with EJ-ras(the activated human c-H-ras) (10a) and are referred to asHPK II-EJras. CaSki, a cervical cancer-derived cell linecontaining about 500 copies of HPV 16 DNA integrated atseveral chromosomal locations (20), was obtained from theAmerican Type Culture Collection.

All cell lines were grown in Dulbecco modified Eaglemedium supplemented with 5% fetal calf serum in a 37°Cincubator with 5% CO2.

Cyst and tumor growth in nude mice. For each cell line, 107cells per animal were injected subcutaneously into 6-week-old female nude mice (BALB/c nulnu backcross). For HPKIA and HPK II cells, cystic nodules developed, reachingmaximal size 2 to 4 weeks after injection, and then re-gressed. Cysts which had persisted for months were shownto consist solely of dead keratinized cells. CaSki cells andHPK II-EJras cells gave rise to progressively growing tu-mors.

Cysts and tumors were excised at different times afterinjection and frozen in cold (-70°C) 2-methylbutane forsubsequent storage in liquid nitrogen.

Indirect immunofluorescence. Serial cryostat sections (5,um thick) were double stained by indirect immunofluores-cence using antibodies directed against various markers forepithelial cell differentiation as first-step reagents and appro-priate fluorescein isothiocyanate- or biotin-labeled anti-mouse, anti-guinea pig, or anti-rabbit immunoglobulins assecond-step reagents (Fig. 1). Use of the second-step re-

First antibodyPanreactiveCytokeratin 10

InvolucrinFilaggrin

Second antibodyAnti-guinea pig FITCAnti-mouse biotinAnti-rabbit FITCAnti-mouse biotin

FIG. 1. Indirect double immunofluorescence. Biotin was de-tected by a third reaction step using Texas red-labeled streptavidin.FITC, Fluorescein isothiocyanate.

agents alone did not give rise to any significant staining of thehuman components of the biopsy specimens, except for deadcornified cells. For double immunofluorescent staining, se-rial sections were each incubated for 45 min at room tem-perature with both first antibodies, i.e., panreactive andanti-cytokeratin 10 antibodies or antiinvolucrin and antifilag-grin antibodies. After several washes in phosphate-bufferedsaline, the sections were incubated with the appropriatesecond antibodies (see Table 1) for an additional 45 min andwashed again. Specifically bound biotinylated anti-mouseimmunoglobulins could be detected after a further incuba-tion with Texas red-labeled streptavidin complex followedby washing. The sections were then embedded in aquamount (BDH Ltd., Poole, England) and photographed (dou-ble exposures with appropriate filters) under an OlympusAH2 microscope equipped with epifluorescence optics.The antiserum against human involucrin was kindly pro-

vided by Fiona Watt (Kennedy Institute of Rheumatology,London, United Kingdom). A panreactive antiserum di-rected against epidermal and simple keratins was a gift fromProgen, Heidelberg, Federal Republic of Germany. Anti-human cytokeratin 10 (CK 8.60) and antibodies directedagainst filaggrin were purchased from Renner, Dannstadt,Federal Republic of Germany, and Paesel, Frankfurt, Fed-eral Republic of Germany, respectively. All second antibod-ies were purchased from Dianova, Hamburg, Federal Re-public of Germany.RNA-RNA in situ hybridization. Serial cryostat sections

mounted on 3-aminopropyl-triethoxysilane-coated slideswere fixed in 4% paraformaldehyde in 2x SSPE (0.3 MNaCl, 20 mM NaH2PO4, 2 mM EDTA) for 10 to 15 min atroom temperature, digested with proteinase K (0.5 ,uglml)for 10 min at 37°C, and hybridized with strand-specific RNAprobes spanning the URR-E6-E7 open reading frames ofHPV 16 (nucleotide positions 7456 to 880) (30) and thehuman cytokeratin 1 gene (17) as described previously (3).Briefly, radioactively labeled RNA probes were generated inthe transcription vector Bluescribe with a solution contain-ing either T3 or T7 RNA polymerase, 100 ,uCi of [32P]UTP(800 Ci/mmol; Amersham), and 0.25 mM each of the remain-ing precursors as a cold substrate. This procedure yields anRNA with a specific activity of 109 cpm/,ug. After DNasetreatment, the probes were subjected to limited alkalinehydrolysis. The sense orientation of each probe served as anegative control. Hybridization was performed overnight at42°C in a solution containing 50% formamide, 2x SSPE, 10%(wt/vol) dextran sulfate, 10 mM Tris (pH 7.5), 1 x Denhardt'ssolution (0.02% each bovine serum albumin, Ficoll, andpolyvinylpyrrolidone), 500 ,ug of tRNA per ml, 100 ,ug ofherring sperm DNA (sonicated) per ml, 0.1% sodium dode-cyl sulfate (SDS), and 105 cpm of probe per RI. Theseconditions approximately equal Tm - 20°C for RNA-RNA insitu hybridization (7). Sections were washed in 50% form-

VOL. 65, 1991

on February 5, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

798 DURST ET AL.

C-

J. VIROL.

on February 5, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

DIFFERENTIATION-DEPENDENT HPV GENE EXPRESSION 799

t~~~~~~~~~~~

'0

0

6

W- ' ~~~~~~~~~~u .b

co .r.E

*~~~~ ~ ~~~~~~~~~~~~~. °x

=n

U.0u

0>

4-0

C,~~~41C,

IL * _

*'~O C.

co

; ,~~~~~~* i-**1*- ,UU2,n

i ..I*,\ .c

VOL. 65, 1991

on February 5, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

800 DURST ET AL.

4--,

C.01)

04>1

Nl

CDOrL i

(I0-'

4

5 A

...1A

&

#8

~..4*

'At 'I,tt

Fb . AV, j

** 4 y, X * t

. * tD

'At r

*FE E

4i,;.S

.<

3*

it'

S

4* .t.t....

* p

..

..~~~~~ ~ ~~~~~~~~~~~~~~. t.w

.~~~~~~~~~~

0 '1

VI >IdH

q <- X g;lB,|e jia :a: Fi'> . . ' . "|i . .'8;g;.: ! t r .;, 2 : r

i. w ^Dv: Xt . ..... ... .. -} £. .; *.;....... s

,,i; .......0

........K .; ,.

*s: :; :'

ie rp tz?

.1 ; .k' S:,.

lF.MN tra W

J. VIROL.

ill

Xt!A J 9 N I 3

on February 5, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

DIFFERENTIATION-DEPENDENT HPV GENE EXPRESSION 801

amide-2x SSPE-0.1% SDS for 30 min at 50°C, treated withRNase A (50 ,ug/ml) for 30 min at 37°C, and subsequentlywashed in 50% formamide-0.5x SSC (lx SSC is 0.15 MNaCl plus 0.015 M sodium citrate)-0.1% SDS for 30 min atthe same temperature (Tm - 15°C). Slides were dehydratedin graded alcohols containing 300 mM ammonium acetate,dried, and dipped in Kodak NTB 2 emulsion diluted 1:1 in600 mM ammonium acetate. After storage for 1 to 14 days,the slides were developed in Kodak D-19 developer, fixed,and counterstained with hematoxylin and eosin. The sec-tions were examined and photographed under a Zeiss Axio-phot microscope equipped with rotatable bright field anddark field condensers to reveal tissue histology and to enablegood signal visualization, respectively.

RESULTSGrowth and differentiation of HPV 16-immortalized kerati-

nocytes and CaSki cells after subcutaneous injection into nudemice. Despite their apparent unlimited growth potential,HPK cells have retained the ability to reform an orderlystructured and differentiated epidermal tissue when injectedsubcutaneously into nude mice. Cystic structures wereformed when suspensions of HPK cells were injected sub-cutaneously onto the subdermal muscle fascia. Approxi-mately 2 weeks after inoculation the epithelium of thesecysts had developed normal tissue architecture, including astratum granulosum and comeum (Fig. 2; compare panels dand g [cysts] and panel a [normal foreskin]). The livingepithelium eventually degenerated, often leaving behind acyst filled with dead keratinized material. The histologicfindings could be substantiated by analysis of differentiation-specific markers. Basal and suprabasal cell compartmentswere clearly distinguishable (as in the epidermis) by doubleimmunofluorescence using a panreactive antibody specificfor all epithelial cells and a polyclonal antibody specific forcytokeratin 10, which is only expressed in the suprabasallayer of the epidermis (21) (Fig. 2; compare panels e and h[cysts] with panel b [normal foreskin]). Moreover, involu-crin, a major precursor protein of the cross-linked envelopein the stratum corneum (27) and filaggrin, the major compo-nent of keratohyalin granules (9), were detected in the upperliving layers of the cysts (Fig. 2; compare panels f and i[cysts] with panel c [normal foreskin]). In addition, humancytokeratin 1 RNA transcription, which in native foreskinand ectocervical tissue is confined to the suprabasal layer,was seen in the corresponding layers in HPK-induced cystsand tumors (see Fig. 4 and 5). HPK cells maintained these invivo differentiation properties at high passage levels, eventhough differentiation in vitro (stratification and squameformation) showed a marked decrease at earlier passages(11).The potential for differentiation was also retained in pro-

gressively growing tumors induced after subcutaneous inoc-ulation of EJ-ras-transfected HPK II cells into nude mice.Evidence for terminal differentiation and cytokeratin 1 RNAexpression is shown in Fig. 5f, i, and k. The same observa-tions were made for tumors induced by CaSki cells, which,

in contrast to tumors induced by HeLa and SiHa cells (datanot shown), also showed signs of terminal differentiation(see Fig. 51, o, and p).The growth properties of normal foreskin keratinocytes

after subcutaneous injection into nude mice were not reex-amined in our experiments, since they have already beenaddressed by others (5, 34). Boukamp et al. described theformation of a regular epithelium with the expression ofdifferentiation-specific markers after transplantation of nor-mal human keratinocytes from different body sites onto nudemice (5).

Analysis of HPV 16 E6-E7 gene expression in nude mousetumors and human biopsy specimens. HPV 16-specific geneexpression could be detected in all cysts and progressivelygrowing tumors induced after subcutaneous injection ofHPV 16-immortalized cells and an HPV 16-positive cervicalcarcinoma cell line into nude mice by RNA-RNA in situhybridization on tissue sections. In each of the experimen-tally induced tumors, cells positive for viral E6-E7 geneexpression were essentially located in the basal layer or inthe extended basallike layer. Differences, however, could beobserved with respect to the distribution of HPV-positivecells and relative signal intensity. In cysts obtained aftersubcutaneous injection of HPK IA and HPK II cells, signalintensity was always weak and positive cells were unevenlydistributed (Fig. 3). In contrast, in progressively growingtumors induced by HPK II-EJras and CaSki cells, the signalstrength was much more intense (compare Fig. 4n and Sc[cysts] with Fig. 5h and n [tumors]). Moreover signals werepresent throughout the undifferentiated, proliferatively ac-tive cell layer. In normal cervical tissue no HPV 16 E6-E7RNA could be detected (Fig. 4c). However, in low-gradeCIN, signals were confined to cells in the terminally differ-entiated layers of the epithelium, where they occurred inhigh concentrations (Fig. 4h). The observations made for thelow-grade lesions are in remarkable contrast to those forhigh-grade lesions, in which HPV 16 E6-E7 transcriptionoccurs throughout the proliferating epithelial layer, whichfails to show evidence for terminal differentiation (lOa).

Differentiation-dependent expression of the viral E6-E7genes. Serial sections of all biopsy samples were hybridizedwith probes for HPV 16 URR-E6-E7 and human cytokeratin1. Cytokeratin 1 is normally expressed in suprabasal cells,which are committed to terminal differentiation (Fig. 4e). Itwas of particular interest to note an inverse relationshipbetween HPV 16 E6-E7 gene expression and that of cyto-keratin 1 in cysts and tumors induced by HPK IA, HPK II,HPK II-EJras, and CaSKi cells (Fig. 4n and p, 5c and e, 5hand k, and 5n and p, respectively; also see Fig. 3 for HPV 16E6-E7 gene expression in HPK IA and HPK II cysts). In allof these lesions, viral RNA transcription was always moreevident in the basallike layer than in differentiated cells.Conversely, in low-grade CIN, HPV 16 E6-E7 transcriptswere mapped to the more differentiated layers of the epithe-lium (Fig. 4h and k).

FIG. 4. RNA-RNA in situ hybridization of sections of normal cervical tissue (a), low-grade CIN (f), and a cyst induced in a nude mouseafter subcutaneous injection ofHPK IA cells (1). (b and c) Light and dark field views, respectively, of a hematoxylin- and eosin-stained sectionof a normal cervix after hybridization with the HPV 16 URR-E6-E7 probe. (d and e) Serial section of the same biopsy specimen hybridizedwith human cytokeratin 1 as a probe. The arrangement of photomicrographs for the other biopsy specimens is the same as that for normalcervical tissue. Specific hybridization signals are seen as white spots by dark field microscopy and, if present in high concentrations, also asblack gains by light field microscopy. Magnification, panels a, f, and 1 x70; other panels x 140.

VOL. 65, 1991

on February 5, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

802 DUJRST ET AL.

K,

Ir_

___1-

>1 i:~~~~~~~~~~~~~~~~~...

Y 4

J. VIROL.

.AA.10r; i

:. A,A..: l:$;'!~~4rl-_

1.Qc)

L.-

:3Qc)T--

>

mi

T-

S| SLis 'wXg x , .X

.1 } 4*wt FL%ss - Xe M

w ks q 1 > tf oAit ';3 t 9 ;>g 8^ > # & . v > A *, * * t ,* * i s s v " "4 4 ; {Hi .et ^ ^ S ^

<t,<* >*,+vsto

8 ,.. .%, _ h, o s ff* .,,,. ,< _ *. * Ht''V.'%<~~~~~~~~~~~~~*

t~~~~l HAar ;r * Bv-,7, * iiKw 45w>=>t \ b: < @1F.

s i > r b

$ ; 4 s + 4 w 9 tA . w f s _ v , 6 i , {tt11 MdH

CJ}} f , t t ^; 5

* 0 s t

;b @ ffi sW ; -. @^+

,, @ * 41

.4e a*r -W .. @

*o s:l * *:

{ ..

4 s*s *.S < *Sf ¢ s j i

J; . S>. w * Xb a qf

*. s

on February 5, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

DIFFERENTIATION-DEPENDENT HPV GENE EXPRESSION 803

DISCUSSION

Immortalization of human keratinocytes in vitro appears

to be a specific property of cervical cancer-linked HPV typessuch as HPV 16, HPV 18, HPV 31, and HPV 33 but not ofHPV 6 and HPV 11, which preferentially induce benigngenital lesions (24, 29, 33). None of the HPV-containing celllines established from primary keratinocytes in differentlaboratories were tumorigenic in nude mice, but they appear

to exhibit a morphology similar to that of CIN when grown

in vitro on collagen rafts which permit epithelial differentia-tion analogous to that seen in vivo (16, 18). Dysplasticdifferentiation in vivo, characteristic of CIN, was also ob-served after grafting HPV-immortalized cell lines beneathskin-muscle flaps of nude mice (34). In contrast, the epider-mal architecture of the transitory cystic nodules induced bysubcutaneous inoculation of two of our HPV 16-immortal-ized keratinocyte cell lines (HPK IA and HPK II) into nudemice is more reminiscent of normal skin than of CIN. Theapparent loss of some differentiation properties in vitro (thatis, the loss of both stratification and squame formation whenthe cells are grown on plastic), which was noticed at an earlystage in the passage history of the HPK cell lines, has notinfluenced the potential for terminal differentiation whenthese cells are grown under in vivo conditions. Since thesetwo cell lines have retained this ability for differentiation invivo despite being in culture for over 2 years (>100 pas-

sages), it is concluded that immortalization of human kera-tinocytes with HPV 16 is not necessarily accompanied bymajor defects in differentiation. There is no obvious expla-nation for the observed differences in the differentiationpotential in vivo of the various HPV-immortalized cell lines(34). Differences in experimental design and conditions aswell as factors unrelated to HPV gene expression (such as

natural variability among different foreskin keratinocytecultures or genetic alterations due to passaging) are likely toinfluence the outcome of each experiment. In contrast to thecell lines examined by Woodworth et al. (34), our cell lineswere established on the basis of continuous passage, i.e.,immortalization after HPV 16 DNA transfection without theuse of any additional dominant selection marker. Moreover,our cultures were continuously grown in a medium with ahigh calcium content which permits epithelial cell differen-tiation. It is even more difficult to reconcile our data withthose obtained for HPV-immortalized cell lines in the raftsystem (16, 18). Although the raft system mimics the in vivosituation to some extent, i.e., cultured cells derived fromvarious tissue biopsy specimens can duplicate many of thehistologic features observed in either normal epithelium orcervical neoplasia (26), complete normalization of the tissuein the raft system is apparently not achieved, as indicated byirregularities in the spatial organization of differentiationproducts (1, 2, 14).By RNA-RNA in situ hybridization, the topographic dis-

tribution of HPV 16-specific RNA in HPK cells exhibitingdifferent degrees of growth potential in nude mice could beassessed. In cysts induced by parental HPK cells, HPV 16E6-E7-specific RNA signals were heterogeneous and were

restricted to cells in the basal layer. However, in progres-sively growing tumors induced by HPK cells containing theactivated human EJ-ras oncogene, high steady-state levelsof viral RNA could be detected throughout the basallikelayer. Northern blot analysis ofRNA extracted from benignand tumorigenic HPK cells grown in vitro showed no quan-titative differences in HPV 16 E6-E7 transcription (data notshown). Thus, HPV 16 gene expression in benign nudemouse tumors, unlike that in cells cultured in vitro, may begoverned by the same cellular factors which are thought tonegatively regulate HPV 18 E6-E7 gene expression in non-tumorigenic HeLa-fibroblast hybrid cell lines (4). Indeed, thevery low levels ofHPV 16 E6-E7-specific RNA signals in thebasal lining of two low-grade CINs which were analyzed atthe same time support a postulated tight cellular control onviral gene expression in actively proliferating cells (4, 35).The mechanism by which the activated Ras protein inter-feres with the postulated modulation of viral gene expressionin vivo is not known.The essence of these results is the observed retention of

the differentiation potential of HPV 16-immortalized kerati-nocytes when analyzed in vivo (nude mouse system) and theinverse relationship between HPV 16 E6-E7 transcriptionand epithelial cell differentiation. In low-grade CIN, HPV 16E6-E7 gene expression is typically confined to the upperlayer of the epithelium and is not or only rarely detected inthe basal layer (8, 31). In our model system, two HPV16-immortalized keratinocyte cell lines, although capable ofterminal differentiation in vivo, are unable to express theseviral genes in the appropriate (with reference to naturallyoccurring lesions) epidermal compartment. Instead, limitedviral transcriptional activity takes place in the basallike layeronly. The same inverse relationship between HPV geneexpression and differentiation was also observed for progres-sively growing tumors induced by EJ-ras-transfected cells(HPK II-EJras) and tumors induced by the cervical carci-noma cell line CaSki.We suspect that the inability to express HPV genes in

differentiated cells may be due to the physical state of theviral DNA within the host cell. In contrast to low-grade CIN,the examined cell lines harbor exclusively integrated HPV 16DNA. Because the viral DNA in cell lines is an integral partof the chromatin, its expression may be regulated differentlyfrom that of episomal DNA. Alternatively, HPV DNAreplication which leads to a high number of DNA templatesalso available for transcription should be considered as apossible explanation for the high level of viral RNA in theupper layer of low-grade lesions. In contrast to naturallyoccurring lesions, in which HPV DNA is known to replicatein the stratum granulosum, none of the nude mouse tumorsinduced by the HPV 16-containing cell lines can supportextrachromosomal viral DNA replication. We are presentlyscreening for well-differentiated HPV 16-positive cervicalcarcinomas which contain only integrated viral DNA. Acomparative analysis of the topographical distribution ofcytokeratin 1 RNA and HPV 16 E6-E7 RNA in these tumorsshould support the observations presented in this paper.

FIG. 5. RNA-RNA in situ hybridization of tissue sections of cysts and tumors induced in nude mice after subcutaneous injection of HPK11 (a), HPK II-EJras (f), and CaSki cells (1). (b and c) Light and dark field views, respectively, of a hematoxylin- and eosin-stained sectionof an HPK II cyst after hybridization with the HPV 16 URR-E6-E7 probe. (d and e) Serial section of the same biopsy specimen hybridizedwith human cytokeratin 1 as a probe. The arrangement of photomicrographs for the other biopsy specimens is the same as that for the HPKII cyst. Specific hybridization signals are seen as white spots by dark field microscopy and, if present in high concentrations, also as blackgrains by light field microscopy. Magnification, panels a, f, and 1 x70; other panels x 140.

VOL. 65, 1991

on February 5, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

804 DURST ET AL.

ACKNOWLEDGMENTS

We are grateful to P. Boukamp and D. Breitkreutz for advice onimmunofluorescent staining and for providing antisera. The gift ofantiserum against human involucrin prepared by F. Watt is greatlyappreciated. We also thank L. Gissmann and colleagues for valuablediscussion.

This work was supported by the Deutsche Forschungsgemein-schaft (grant D0162/1-1).

REFERENCES1. Asselineau, D., B. A. Bernard, C. Bailly, M. Darmon, and M.

Prunieras. 1986. Human epidermis reconstituted by culture: it is"normal"? J. Invest. Dermatol. 86:181-186.

2. Bohnert, A., J. Hornung, I. C. Mackenzie, and N. E. Fusenig.1986. Epithelial-mesenchymal interactions control basementmembrane production and differentiation in cultured and trans-planted mouse keratinocytes. Cell Tissue Res. 244:413-429.

3. Bosch, F. X., J.-P. Ouhayoun, B. L. Bader, C. Collin, C. Grund,I. Lee, and W. W. Franke. 1989. Extensive changes in cytoker-atin expression patterns in pathologically affected human gin-giva. Virchows Arch. B Cell. Pathol. 58:59-77.

4. Bosch, F. X., E. Schwarz, P. Boukamp, N. E. Fusenig, D.Bartsch, and H. zur Hausen. 1990. Suppression in vivo of humanpapillomavirus type 18 E6-E7 gene expression in nontumori-genic HeLa x fibroblast hybrid cells. J. Virol. 64:4743-4754.

5. Boukamp, P., D. Breitkreutz, H.-J. Stark, and N. E. Fusenig.1990. Mesenchyme-mediated and endogenous regulation ofgrowth and differentiation of human skin keratinocytes derivedfrom different body sites. Differentiation 44:150-161.

6. Buckley, C. H., E. B. Butler, and H. Fox. 1982. Cervicalintraepithelial neoplasia. J. Clin. Pathol. 35:1-13.

7. Cox, K. H., D. V. DeLeon, L. M. Angerer, and R. C. Angerer.1984. Detection of mRNAs in sea urchin embryos by in situhybridization using asymmetric RNA probes. Dev. Biol. 101:485-502.

8. Crum, C. P., G. Nuovo, D. Friedman, and S. J. Silverstein. 1988.Accumulation of RNA homologous to human papillomavirustype 16 open reading frames in genital precancers. J. Virol.62:84-90.

9. Dale, B. A., K. A. Resing, P. V. Haydock, P. Fleckman, C.Fisher, and K. Holbrook. 1989. Intermediate filament associatedprotein of epidermis, p. 97-115. In G. E. Rogers, P. J. Reis,K. A. Ward, and R. C. Marshall (ed.), The biology of wool andhair. Chapman & Hall, Ltd., London.

10. DiPaolo, J. A., C. D. Woodworth, N. C., Popescu, V. Notario,and J. Doniger. 1989. Induction of human cervical squamouscell carcinoma by sequential transfection with human papillo-mavirus 16 DNA and viral Harvey ras. Oncogene 4:395-399.

10a.Durst, M., et al. Unpublished data.11. Durst, M., R. T. Dzarlieva-Petrusevska, P. Boukamp, N. E.

Fusenig, and L. Gissmann. 1987. Molecular and cytogeneticanalysis of immortalized human primary keratinocytes obtainedafter transfection with human papillomavirus type 16 DNA.Oncogene 1:251-256.

12. Durst, M., D. Gallahan, J. Gilbert, and J. S. Rhim. 1989.Glucocorticoid-enhanced neoplastic transformation of humankeratinocytes by human papillomavirus type 16 and an activatedras oncogene. Virology 173:767-771.

13. Dyson, N., P. M. Howley, K. Munger, and E. Harlow. 1989. Thehuman papillomavirus 16 E7 oncoprotein is able to bind to theretinoblastoma gene product. Science 243:934-937.

14. Fusenig, N. E., D. Breitkreutz, P. Boukamp, A. Bohnert, andI. C. Mackenzie. In N. Johnson (ed.), oral cancer: detection ofpatients and lesions at risk, in press. Cambridge UniversityPress, Cambridge.

15. Hawley-Nelson, P., K. H. Vousden, N. L. Hubbert, D. R. Lowy,and J. T. Schiller. 1989. HPV 16 E6 and E7 proteins cooperateto immortalize human foreskin keratinocytes. EMBO J. 8:3905-3910.

16. Hudson, J. B., M. A. Bedell, D. J. McCance, and L. A. Laimins.1990. Immortalization and altered differentiation of human ke-

ratinocytes in vitro by the E6 and E7 open reading frames ofhuman papillomavirus type 18. J. Virol. 64:519-526.

17. Knapp, A. C., and W. W. Franke. 1989. Spontaneous losses ofcontrol of cytokeratin gene expression in transformed, non-epithelial human cells at different levels of regulation. Cell59:67-79.

18. McCance, D. J., R. Koplan, E. Fuchs, and L. A. Laimins. 1988.Human papillomavirus type 16 alters human epithelial celldifferentiation in vitro. Proc. Natl. Acad. Sci. USA 85:7169-7173.

19. Meisels, A., R. Fortin, and M. Roy. 1977. Condylomatous lesionof the cervix. II. Cytologic, colposcopic and histopathologicstudy. Acta Cytol. 21:379-390.

20. Mincheva, A., L. Gissmann, and H. zur Hausen. 1987. Chromo-somal integration sites of human papillomavirus DNA in threecervical cancer cell lines mapped by in situ hybridization. Med.Microbiol. Immunol. 176:245-256.

21. Moll, R., W. W. Franke, and D. L. Schiller. 1982. The catalog ofhuman cytokeratins: pattern of expression in normal epithelia,tumors and cultured cells. Cell 31:11-24.

22. Munger, K., W. C. Phelps, V. Bubb, P. M. Howley, and R.Schlegel. 1989. The E6 and E7 genes of the human papillomavi-rus type 16 together are necessary and sufficient for transforma-tion of primary human keratinocytes. J. Virol. 63:4417-4421.

23. Munger, K., B. A. Werness, N. Dyson, W. C. Phelps, E. Harlow,and P. M. Howley. 1989. Complex formation of human papillo-mavirus E7 proteins with the retinoblastoma tumor suppressorgene product. EMBO J. 8:4099-4105.

24. Pecoraro, G., D. Morgan, and V. Defendi. 1989. Differentialeffects of human papillomavirus type 6, 16, and 18 DNAs onimmortalization and transformation of human cervical epithelialcells. Proc. Natl. Acad. Sci. USA 86:563-567.

25. Pirisi, L., S. Yasumoto, M. Feller, J. Doniger, and J. A. DiPaolo.1987. Transformation of human fibroblasts and keratinocyteswith human papillomavirus type 16 DNA. J. Virol. 61:1061-1066.

26. Rader, J. S., T. R. Golub, J. B. Hudson, D. Patel, M. A. Bedell,and L. A. Laimins. 1990. In vitro differentiation of epithelialcells from cervical neoplasias resembles in vivo lesions. Onco-gene 5:571-576.

27. Rice, R. H., and H. Green. 1979. Presence in human epidermalcells of a soluble protein precursor of the cross-linked envelope:activation of the cross-linking by calcium ions. Cell 18:681-694.

28. Richart, R. M. 1973. Cervical intraepithelial neoplasia: a re-view. Pathol. Annu. 8:301-328.

29. Schlegel, R., W. C. Phelps, Y.-L. Zhang, and M. Barbosa. 1988.Quantitative keratinocyte assay detects two biological activitiesof human papillomavirus DNA and identifies viral types associ-ated with cervical carcinoma. EMBO J. 7:3181-3187.

30. Seedorf, K., G. Krammer, M. Durst, S. Suhai, and W. Rowe-kamp. 1985. Human papillomavirus type 16 DNA sequence.Virology 145:181-185.

31. Stoler, M. H., C. R. Rhodes, A. Whitbeck, L. T. Chow, andT. R. Broker. 1990. Gene expression of HPV types 16 and 18 incervical neoplasia, p. 1-11. In P. M. Howley and T. R. Broker(ed.), Papillomaviruses. Wiley-Liss, Inc., New York.

32. Werness, B. A., A. J. Levine, and P. M. Howley. 1990. Associ-ation of human papillomavirus type 16 and 18 E6 proteins withp53. Science 248:76-79.

33. Woodworth, C. D., J. Doniger, and J. A. DiPaolo. 1989. Immor-talization of human foreskin keratinocytes by various humanpapillomavirus DNAs corresponds to their association withcervical carcinoma. J. Virol. 63:159-164.

34. Woodworth, C. D., S. Waggoner, W. Barnes, M. H. Stoler, andJ. A. DiPaolo. 1990. Human cervical and foreskin epithelial cellsimmortalized by human papillomavirus DNAs exhibit dysplasticdifferentiation in vivo. Cancer Res. 50:3709-3715.

35. zur Hausen, H. 1984. Intracellular surveillance of persisting viralinfections: human genital cancer results from deficient cellularcontrol of papillomavirus gene expression. Lancet ii:489-491.

36. zur Hausen, H. 1989. Papillomaviruses as carcinomaviruses.Adv. Viral Oncol. 8:1-26.

J. VIROL.

on February 5, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from